Font Format Converter — WOFF2 to TTF, TTF to WOFF/WOFF2 in Your Browser
Convert web fonts between TTF, WOFF and WOFF2 without leaving the page: the source format is detected from the file signature, glyph coverage and TrueType hinting are carried over unchanged, and nothing is uploaded — the WOFF2 codec is a local WebAssembly build. A 20 MB CJK font works but takes tens of seconds, one file at a time.
What this font converter changes — and what it leaves untouched
Three formats in, three formats out
Pick a .ttf, .woff or .woff2 file, choose TTF, WOFF or WOFF2 as the target and download the result. The source format is sniffed from the signature (wOFF, wOF2, true/OTTO/0x00010000), so you never state it yourself; choosing the same format returns the original bytes.
WOFF2 to TTF and back, locally
WOFF2 encoding and decoding run through Google's WOFF2 codec compiled to WebAssembly inside the page, while the older WOFF container is handled by pako in plain JavaScript — there is no server round trip.
Glyph coverage and hinting survive
The tool repackages the font tables instead of re-rendering outlines: glyph count and cmap coverage stay identical, and TrueType hinting tables (cvt, prep, fpgm, glyf instructions) come through untouched. The only table the WOFF2 encoder drops is DSIG, the digital signature.
Built for web-font compatibility work
Ship WOFF2 to modern browsers, keep WOFF or TTF for older stacks, or normalize a mixed font folder to one format. If you only want a smaller file, use the font subsetter instead — this page never removes glyphs.
Limits it is honest about
One file at a time (only the first selected file is read), no weight or italic controls, and TTC collections are rejected as an unrecognized signature. A 20 MB CJK font does convert, but it takes tens of seconds and hundreds of megabytes of RAM.
Font format conversion FAQ
Which input and output formats are supported?
Input: TTF (TrueType sfnt), WOFF and WOFF2 — the three extensions the file picker accepts — and the source format is detected automatically. Output: TTF, WOFF or WOFF2 from the dropdown. TTC/OTC collections fail with 'Unrecognized font signature: ttcf', and OTF is not offered in the picker even though the parser would recognize an OTTO/CFF font.
Can I adjust weight or make the font italic?
No. Conversion only rewraps the existing outlines and tables; it does not synthesize bold, slant or extra weights. Those need a type-design tool or a variable-font axis, and this page has no such controls.
Is character coverage or hinting lost in the conversion?
Coverage and TrueType hinting are preserved: every table is copied across, glyph counts stay identical (a 20.9 MB Song font kept all 60,455 glyphs), and cvt, prep, fpgm plus glyf instructions remain. WOFF2 encoding drops only the DSIG signature table; WOFF output keeps every table.
Can it handle large CJK fonts, and how much memory does it need?
Yes, but there is no size cap and the whole file is buffered in memory. Measured here: a 20.9 MB TTF became an 8.6 MB WOFF2 in roughly 55 seconds with about 350 MB of memory; a 16.9 MB Noto Sans SC took about 39 seconds at about 400 MB. The main thread is blocked meanwhile, so the page freezes until it finishes — subset the font first when you can.
Can I convert several fonts in one batch?
No. The picker reads a single file (files[0]) and each run overwrites the previous result; there is no queue, progress list or drag-and-drop batch mode.
Is this local or does it use a server?
Entirely local. WOFF2 encode and decode run in an Emscripten WebAssembly module embedded as a data URI, and WOFF wrapping uses pako in JavaScript. The font is never uploaded, so licensed fonts stay private.
What happens if conversion fails, and why is the file always named font.*?
If conversion throws, nothing is downloaded and the error is only logged to the console; the success toast appears after the blob has been built. The download name is hard-coded as font.ttf, font.woff or font.woff2, so rename it to match your font family before shipping.

